Skip to main content
A simple email agent demonstrating integration with Google’s Agent Development Kit (ADK) and Resend API. Shows how to structure agents with external API tools and run them via ADK CLI.

​
Features

  • Built with Google ADK framework
  • Email sending via Resend API
  • LiteLLM model integration
  • ADK CLI and Dev UI support
  • Production-ready structure

​
Prerequisites

​
Installation

1

Clone the repository

git clone https://github.com/Arindam200/awesome-ai-apps.git
cd starter_ai_agents/google_adk_starter
2

Install dependencies

pip install -r requirements.txt
Verify ADK installation:
adk --version
3

Configure environment

Copy the example environment file:
cp .env.example .env
Edit .env with your API keys:
NEBIUS_API_KEY="your_nebius_api_key_here"
NEBIUS_API_BASE="https://api.studio.nebius.ai/v1"
RESEND_API_KEY="your_resend_api_key_here"

​
Implementation

​
Email Tool

Create a tool for sending emails:
agent.py
from dotenv import load_dotenv
import os
import resend
from google.adk.agents import Agent
from google.adk.models.lite_llm import LiteLlm

load_dotenv()
resend.api_key = os.getenv("RESEND_API_KEY")

def send_email(_: str) -> dict:
    """
    Sends an email using the Resend API.

    Args:
        _: Placeholder argument (not used).

    Returns:
        dict: A dictionary containing the status and message ID or error message.
    """
    try:
        params = {
            "from": "Anand Demo <[email protected]>",  # Replace with your verified sender
            "to": ["[email protected]"],     # Replace with recipient
            "subject": "Test Email from Google ADK",
            "html": "<p>This is a test email sent from a Google ADK agent using Resend.</p>",
        }
        response = resend.Emails.send(params)
        return {"status": "success", "message_id": response.get("id")}
    except Exception as e:
        return {"status": "error", "message": str(e)}

​
Model Configuration

agent.py
# Initialize the LiteLlm model
model = LiteLlm(
    model="openai/meta-llama/Meta-Llama-3.1-8B-Instruct",
    api_base=os.getenv("NEBIUS_API_BASE"),
    api_key=os.getenv("NEBIUS_API_KEY")
)

​
Agent Definition

agent.py
# Define the EmailAgent
root_agent = Agent(
    name="EmailAgent",
    model=model,
    description="Agent that sends emails using the Resend API.",
    instruction="Use the send_email tool to send an email.",
    tools=[send_email]
)

# Execute the email sending function
result = send_email("")
print(result)

​
Usage

​
Run with ADK CLI

# Run directly in the terminal
adk run email_adk_agent

​
Run Directly

python agent.py

​
Technical Details

​
Google ADK Components

Agent

Core agent class with name, model, and tools

LiteLlm

Model provider supporting multiple LLM backends

Tools

Python functions exposed to the agent

ADK CLI

Command-line interface for running agents

​
Agent Configuration

Agent(
    name="EmailAgent",              # Agent identifier
    model=model,                    # LLM model instance
    description="...",              # What the agent does
    instruction="...",              # How to use tools
    tools=[send_email]              # Available functions
)

​
ADK CLI Commands

​
Development

# Start dev UI
adk web

# Run in terminal
adk run email_adk_agent

# Check version
adk --version

# Get help
adk --help

​
Agent Testing

The ADK Dev UI provides:
  • Interactive chat interface
  • Tool execution visualization
  • Debug logs
  • Request/response inspection

​
Extending the Agent

​
Add More Tools

def fetch_emails(_: str) -> dict:
    """Fetch recent emails"""
    # Implementation
    return emails

def schedule_email(recipient: str, subject: str, body: str, send_at: str) -> dict:
    """Schedule an email for later"""
    # Implementation
    return {"status": "scheduled", "send_at": send_at}

root_agent = Agent(
    name="EmailAgent",
    model=model,
    description="Advanced email agent with scheduling",
    instruction="Use the available tools to send, fetch, and schedule emails.",
    tools=[send_email, fetch_emails, schedule_email]
)

​
Customize Email Content

def send_custom_email(recipient: str, subject: str, body: str) -> dict:
    """
    Send a customized email.
    
    Args:
        recipient: Email address of recipient
        subject: Email subject line
        body: HTML email body
    """
    try:
        params = {
            "from": "Your Name <[email protected]>",
            "to": [recipient],
            "subject": subject,
            "html": body,
        }
        response = resend.Emails.send(params)
        return {"status": "success", "message_id": response.get("id")}
    except Exception as e:
        return {"status": "error", "message": str(e)}

​
Use Different Models

# Use a different Nebius model
model = LiteLlm(
    model="openai/Qwen/Qwen3-30B-A3B",
    api_base=os.getenv("NEBIUS_API_BASE"),
    api_key=os.getenv("NEBIUS_API_KEY")
)

​
Add Multi-Agent System

email_agent = Agent(
    name="EmailAgent",
    model=model,
    description="Sends emails",
    tools=[send_email]
)

scheduler_agent = Agent(
    name="SchedulerAgent",
    model=model,
    description="Schedules tasks",
    tools=[schedule_task]
)

coordinator = Agent(
    name="Coordinator",
    model=model,
    description="Coordinates email agent and scheduler",
    sub_agents=[email_agent, scheduler_agent]
)

​
Best Practices

  • Keep tools focused and simple
  • Provide clear docstrings
  • Handle errors gracefully
  • Return structured responses
  • Write clear descriptions
  • Provide specific instructions
  • Use descriptive names
  • Test with ADK Dev UI
  • Use verified sender addresses
  • Validate recipient emails
  • Handle API rate limits
  • Log email status

​
Error Handling

def send_email_safe(recipient: str, subject: str, body: str) -> dict:
    """Send email with comprehensive error handling"""
    try:
        # Validate inputs
        if not recipient or '@' not in recipient:
            return {"status": "error", "message": "Invalid recipient email"}
        
        # Send email
        params = {
            "from": "Your Name <[email protected]>",
            "to": [recipient],
            "subject": subject,
            "html": body,
        }
        response = resend.Emails.send(params)
        
        return {
            "status": "success", 
            "message_id": response.get("id"),
            "recipient": recipient
        }
    
    except resend.exceptions.APIError as e:
        return {"status": "error", "message": f"API Error: {str(e)}"}
    except Exception as e:
        return {"status": "error", "message": f"Unexpected error: {str(e)}"}

​
Environment Variables

VariableDescriptionDefaultRequired
NEBIUS_API_KEYNebius API key-Yes
NEBIUS_API_BASENebius API endpointhttps://api.studio.nebius.ai/v1Yes
RESEND_API_KEYResend API key-Yes

​
ADK Dev UI Features

  • Chat Interface: Test agent interactions
  • Tool Visualization: See tool calls in real-time
  • Debug Console: View logs and errors
  • Request Inspector: Examine LLM requests/responses
  • Agent Switcher: Test multiple agents

​
Next Steps

Advanced ADK

Build complex multi-agent systems

Tool Integration

Connect more external services

Build docs developers (and LLMs) love

Get started for freeTalk to us